Aprendizado de máquina no controle de qualidade na indústria farmacêutica e alimentícia
Nenhuma Miniatura disponível
Data
2019-11-12
Autores
Lima, Felipe da Silva [UNIFESP]
Orientadores
Lopes, Patricia Santos [UNIFESP]
Tipo
Trabalho de conclusão de curso de graduação
Título da Revista
ISSN da Revista
Título de Volume
Resumo
O Controle de Qualidade é responsável pelas atividades referentes à amostragem, às especificações e aos ensaios de produtos acabados e de matérias primas, tendo vital importância nas indústrias farmacêuticas e alimentícias por estas lidarem diretamente com a saúde dos consumidores finais. Aprendizado de máquina é um subcampo de estudo da área de Inteligência Artificial que lida com agentes/algoritmos que podem melhorar o seu comportamento ou desempenho através dos estudos de suas próprias experiências prévias. Trabalhos recentes visam juntar os dois conceitos, apresentando novas técnicas e metodologias para melhorar em diversos aspectos o controle de qualidade nas indústrias farmacêuticas e alimentícias. Foram realizadas buscas de artigos nas bases de dados Scopus e Web of Science, utilizando os descritores quality control, machine learning, pharmaceutical, food e bacterial sendo identificados 717 estudos, dos quais foram incluídos 16 artigos completos. Desses estudos, 10 focam em análise de medicamentos e matérias primas e 6 focam em análises de alimentos. A literatura abrange diversas técnicas analíticas com diversos algoritmos de aprendizado, sendo os mais vistos a Espectroscopia Raman e FT-IR usando principalmente os algoritmos de Support Vector Machine e Redes Neurais Artificiais. Apesar de apresentarem bons dados estatísticos e baixo custo de aplicação, é de senso comum dos autores que ainda precisam ser feitos estudos mais aprofundados para tornar esses métodos realmente úteis e viáveis.
Quality Control is the department responsible for the activities such as sampling, specifications and test the final products and raw material, having a vital importance at the pharmaceutical and food industries, which deal directly with the health of the final customers. Machine learning is a subfield of studies in the Artificial Intelligence field, which works with agents/algorithms that can improve their behavior or performance by studying their own previous experiences. Recent studies aim at place together both concepts, presenting us new techniques and methodologies to improve the quality control in the pharmaceutical and food industries. It was searched articles at the Scopus and Web of Science databases, using the descriptors quality control, machine learning, pharmaceutical, food and bacterial resulting in 717 articles, which 16 have been fully included in this work. Of these articles, 10 focus on drugs and raw material analysis, and 6 focus on food analysis. The literature covers several analytical techniques associated with several learning algorithms, which the most seen are Raman Spectroscopy and FT-IR using the Support Vector Machine and Artificial Neural Network algorithms. Despite of showing good statistic data and low cost application, the authors agree that still need more in-depth studies to render these methods really useful and viable.
Quality Control is the department responsible for the activities such as sampling, specifications and test the final products and raw material, having a vital importance at the pharmaceutical and food industries, which deal directly with the health of the final customers. Machine learning is a subfield of studies in the Artificial Intelligence field, which works with agents/algorithms that can improve their behavior or performance by studying their own previous experiences. Recent studies aim at place together both concepts, presenting us new techniques and methodologies to improve the quality control in the pharmaceutical and food industries. It was searched articles at the Scopus and Web of Science databases, using the descriptors quality control, machine learning, pharmaceutical, food and bacterial resulting in 717 articles, which 16 have been fully included in this work. Of these articles, 10 focus on drugs and raw material analysis, and 6 focus on food analysis. The literature covers several analytical techniques associated with several learning algorithms, which the most seen are Raman Spectroscopy and FT-IR using the Support Vector Machine and Artificial Neural Network algorithms. Despite of showing good statistic data and low cost application, the authors agree that still need more in-depth studies to render these methods really useful and viable.